I've just finished a series of interactive pieces for a training course using the HTML5 Canvas document type in Flash CC. Mostly, things work great, but we've had to convert all of our text to outlines or bitmaps due to the unpredictable layout and other limitations of "Dynamic Text". This time around, we've got recorded audio to read all the text out loud, but we're looking for other solutions for screen-readers to be able to identify and read text in future HTML5 Canvas interactions. Can they read Dynamic Text? Is there a better way to circumvent the limitations of Dynamic Text? Or perhaps there are entirely separate solutions that others have found for facilitating accessibility?